When English Rewrites Local Knowledge: Global Narrative Dominance in Large Language Models

2026-05-28

AI Summary

This study addresses the susceptibility of large language models to dominant global narratives in cross-lingual question answering, which often undermines local cultural expression. The authors introduce CulturalNB, the first Bengali–English parallel cultural QA dataset comprising 717 instances with sociocultural annotations, and systematically evaluate nine mainstream models under varying query languages and evidence-prompting conditions. Employing both question-only and evidence-augmented prompting strategies, and leveraging human and dual-LLM evaluators, the analysis—assessed along dimensions of cross-lingual consistency, linguistic anchoring, and global substitution—reveals for the first time a language-induced cognitive bias: English queries significantly amplify reliance on global narratives and institutionalized frameworks while diminishing coverage of local perspectives. Although locally grounded evidence improves factual consistency, it proves insufficient to fully mitigate this bias.

Abstract

Large language models (LLMs) are widely used as cross-lingual knowledge interfaces. However, culturally grounded questions often reflect globally dominant narratives rather than local contexts. We study this failure mode as \textit{global narrative dominance} in Bangla, a low-resource cultural context. We introduce \texttt{CulturalNB}, a dataset of 717 manually curated Bengali cultural instances with parallel Bangla--English question--answer pairs and supporting evidence, metadata, and sociocultural annotations. Using question-only and evidence-based prompting, we evaluate nine state-of-the-art LLMs with human and two independent LLM judges across metrics for cross-lingual consistency, language anchoring, global substitution, institutional bias, and epistemic perspective coverage. Results show that questions asked in English systematically increase global substitution and institutional framing while reducing local perspective coverage. Local evidence improves factual consistency and perspective coverage, but does not eliminate language-induced epistemic shifts. These findings suggest that cultural failures in LLMs are not only missing-knowledge errors but also failures of grounding and narrative prioritization.

Problem

• global narrative dominance
• cross-lingual knowledge
• cultural grounding
• local knowledge
• language models

Innovation

• global narrative dominance
• cross-lingual grounding
• CulturalNB dataset
• language-induced epistemic shift
• local knowledge representation